it takes two to tango: couples' happiness and …1 it takes two to tango: couples'...
TRANSCRIPT
WP 15
The research leading to these results has received funding from the European Research Council under the European ERC Grant Agreement n. StG-313617 (SWELL-FER: Subjective Well-being and Fertility): PI. Letizia Mencarini
It takes two to tango: couples' happiness and childbearing
Arnstein Aassvea Bruno Arpinob
Nicoletta Balboa
a) Dondena Centre for Research on Social Dynamics, Bocconi University
b) Universitat Pompeu Fabra
1
It takes two to tango: couples' happiness and childbearing
Arnstein Aassve1, Bruno Arpino2, Nicoletta Balbo1
1 - Department of Policy Analysis and Public Management and Carlo F. Dondena Centre for Research on
Social Dynamics and Public Policies, Bocconi University, Milan, Italy; [email protected];
2 - Department of Political and Social Sciences and the Research and Expertise Centre for Survey
Methodology (RECSM), Pompeu Fabra University, Barcelona, Spain; [email protected].
Abstract
Existing literature has so far considered the role of the individual's subjective well-being
on fertility, neglecting the importance of the partner’s well-being. Using data from the
British Household Panel Survey (BHPS) and event history models estimated separately
by parity, we find that in a couple, women's happiness matter more than that of the male
partner in terms of having the first child. Specifically, we observe that couples in which
either partner is happier are more at risk of having the first child but the effect is
strongest with higher happiness of the woman. For the transition to the second child we
find that couples in which the woman is either happier or less happy than usual, are
associated with a lower risk of childbirth. We moreover find support for a multiplicative
effect of partners’ SWB on the decision to have a first child. Our results show that
failing to acknowledge that the subjective well-being of both partners matter for the
inherently joint decision making of childbearing, can lead to an uncompleted view of
how subjective well-being affects fertility.
2
1. Introduction
No longer limited to psychology, studies on subjective wellbeing and the way it links
with various behavioral aspects, are finding their way into the social sciences. Within
demography, there has been a particular focus on the relationship between subjective
wellbeing and childbearing. Recent years have witnessed a large number of studies that
consider the dynamic interplay between childbearing and subjective wellbeing (SWB)
(e.g., Billari 2009; Aassve et al., 2012; Margolis and Myrskylä 2011, 2015). Although
the majority of them focuses on the effect of fertility on SWB, others investigate the
opposite relationship, that is, the role of SWB on intended and actual fertility.
Existing theories and findings are rather mixed. Higher life satisfaction has been found
to predict intended (Billari 2009; Perelli-Harris 2006) and actual fertility (Parr 2010).
More recent studies have showed that an individual’s SWB is particularly relevant for
having a second child, because such decision is very much a function of the experience
of having had the first child - and importantly - the satisfaction associated with it (Le
Moglie 2015; Margolis and Myrskylä 2015). Therefore those individuals who
experience a higher life satisfaction (or a lower drop) after the first child birth are more
likely to have another one. Conversely, Mc Donald (2002) has highlighted how happier
people may refrain from having children in order not to change a positive status-quo.
Such risk aversion have been used as a possible explanation for the rise in voluntary
childlessness (Mencarini and Tanturri 2007).
The vast majority of the studies on SWB and childbearing focuses on individuals’, or
rather, the respondents’ wellbeing (e.g., Myrskylä and Margolis 2014). This is perhaps
surprising because the venture of childbearing is necessarily a joint decision of the two
partners involved (Bauer and Kneip 2013). In this paper we ask the question to what
3
extent a couple-perspective brings insights into the relationship between subjective
wellbeing and couples’ childbearing decision making. Consequently, childbearing
events constitute the dependent variable, and the key explanatory variable refers to the
reported level of happiness of both partners. The analysis is then implemented by parity,
acknowledging that potential coherency or mismatch between the two partners may
have different impact when considering becoming a parent first time, as opposed to the
decision to have another child. Of particular interest is to understand to what extent a
potential mismatch in subjective wellbeing across partners may affect their decision
making. Similarly, interest lies in understanding to what extent there are multiplicative
effects. That is, can one detect an alleviated effect on childbearing if both partners are
closely in line when it comes to their reported subjective wellbeing?
The analysis is based on the British Household Panel Survey (BHPS), from which we
observe fertility behaviors and subjective well-being of couples over a period of 18
years (from 1991 to 2008). We implement a series of event history models to investigate
whether the level of happiness of the two partners considered together, affects fertility.
2. Background
There is now a growing body of studies considering the relationship between fertility
and subjective wellbeing. Compared to earlier studies of fertility within the field of
demography, this line of analysis represents a considerable shift. But given the
introduction of the Second Demographic Transition (SDT) some 40 years ago, the
paradigm of subjective wellbeing comes naturally when considering childbearing
behavior. Inspired by the rise of Post-Materialism (Inglehart 1971), the main idea of the
4
SDT is that the family has become less essential (Van de Kaa 1987), and so new
demographic behavior are emerging, which would include divorce, cohabitation and
out-of-wedlock childbearing and, importantly, fertility postponement and decline
(Aassve et al. 2013). Implicit in the SDT lies the idea that individuals' value orientation
are changing with an increasing emphasis on freedom of expression and, importantly,
psychological wellbeing. Individuals are in a continuous quest for improving their
subjective wellbeing (Le Moglie et al. 2015), but given the ever increasing complexity
of individuals' lives, obtaining fulfillment through children becomes necessarily only
one element out of many. In other words, the wellbeing associated with childbearing,
increasingly depends on the timing, context and the way it is compatible with a range of
other activities that individuals now give high priority.
There is consequently no surprise that in recent years a series of studies
analyzing the relationship between happiness and childbearing has emerged (Aassve et
al. 2012, 2015; Balbo and Arpino 2014; Baranowska and Matysiak 2011; Kohler et al.
2005; Margolis and Myrskylä 2011, 2015; Myrskylä and Margolis 2014). Subjective
wellbeing is usually proxied either by a measure of life satisfaction, or more frequently,
the reported level of happiness, and held up against childbearing behavior.
Most of these studies focus on how individuals associate childbearing with something
positive and aim at uncover whether having children affect SWB. In other words, in the
empirical analysis, the dependent variable typically refers to the reported level of SWB,
and childbearing events are taken as the key explanatory variables. However, one
frequently observed pattern is that SWB increases prior to childbearing, whereas after
the childbearing event there is a great deal of adaption, and often the positive
anticipation effect is subsequently neutralized (Balbo and Arpino 2014; Clark et al.
5
2008; Myrskylä and Margolis 2014). The very fact that there is a significant anticipation
effect, has prompted interest to understand to what extent an increase in SWB leads to
an increase in the probability of childbearing. Some studies have therefore investigated
how the level of happiness or life satisfaction prior to childbearing affects the
subsequent decision of having a child. Billari (2009), for instance, finds that happier
people are more likely to intend to have a(nother) child. Consistent results are found by
Perelli-Harris (2006), who finds the same positive effect of SWB on intended and actual
fertility in Russia. Parr (2010), using longitudinal data, finds a significant positive
relationship between life satisfaction and subsequent fertility in Australia. Parr elaborate
on the possible mechanisms according to which SWB would affects fertility, net of
other, relevant socio-demographic factors (e.g., employment, income). First, the
presence of a partner would contribute to an individual’s life satisfaction as well as to
fertility. Moreover, a satisfying partnership (which would positively affect an
individual’s SWB) would increase the likelihood of having a child for the two partners
through contributing to the stability and the likelihood of the union itself. Third, because
parents desire to have happy children, happier parents may increase the likelihood that
their children will become so, thereby increasing also their likelihood to have one.
Conversely, depression and stress have been found to reduce fecundity and in turn
fertility. Finally, as Kohler et al. (2005) stated, the satisfaction deriving from having
children contributes to the overall SWB of the parents, therefore, the parental
experience and relative satisfaction may affect further parents’ fertility. Based on this
argument, Margolis and Myrskylä (2015), investigates how the change in life
satisfaction of parents after the birth of the first child influences the subsequent decision
6
to have another one. They find that those parents who experience a lower drop in life
satisfaction after first childbirth are those more likely to progress to parity 2.
Another relevant study is the one of Le Moglie et al. (2015). They use the German
Socio Economic Panel Survey to analyze how SWB affects the likelihood of having
children. Their study puts a particular focus on the way personality traits interacts with
both SWB and childbearing, and importantly, they do so separately by parity. They find
that higher SWB leads to a higher likelihood of childbearing, but only for the second
child. For the onset of parenthood, and having the third child, there is no effect.
However, their results are of high importance, because low fertility, other than driven by
childlessness, is in large part explained by lower progression to having the second child.
Moreover, for the progression to the second child, the effect is significant only for
women.
What has so far been neglected in the contributions on the nexus between SWB and
childbearing is that childbearing is necessarily a joint decision between two partners,
which implies that the SWB of both partners should be jointly considered as
determinant of fertility. Almost all of the existing studies take the respondent as the
unit of analysis - holding his or her SWB together with childbearing events. In other
words, if one believes that SWB of the respondent has a direct effect on childbearing
behavior, which is indeed demonstrated by Le Moglie et al. (2015), then intuitively one
would also expect the SWB of the partner to play a role. Exactly how the SWB of the
respondent interacts with the SWB of the partner for childbearing decision making is
not at all clear - nor is it obvious how any such interactions may differ across parity.
7
There is however strong evidence that the reported SWB of the respondent is not
independent of the SWB of the partner. In a recent study, Powdthavee (2009), using the
British Household Panel Survey (BHPS), demonstrates a positive correlation between
spouses' self reported life satisfaction. He also explores the underlying mechanism
behind the correlation, and postulates three key factors. First, individuals with an innate
inclination towards higher life satisfaction may also partner individuals who are similar
in this respect. This follows up on an established literature on assortative mating
(Becker et al. 1977; Greenwood et al. 2014). Second, partnerships allow sharing of
physical and emotional resources that are unavailable if remaining single, and third, any
observed correlation may be a result of direct spillover in SWB within the couple. This
last mechanism refers to the idea that if one partner cares about the other, then the SWB
of the latter becomes a significant driver of the SWB of the former - and vice versa.
Using a dynamic panel model and adjusting for measurement errors, Powdthavee
indeed finds evidence of significant spillover effects.
Of interest in our context however, is to what extent partners' wellbeing may
affect objective measures of demographic behavior, and specifically the decision of
having a child. There is a large literature demonstrating that dissimilarity in partner's
characteristics tend to affect marital stability, the argument being that dissimilarity
associates positively towards marital disruption (Jalovaara 2003; Clarkwest 2007;
Milewski and Kulu 2014). When it comes to assessing the effect of dissimilarity
measured in terms of SWB, the literature is less developed, but there are exceptions.
Guven et al. (2012) use longitudinal data from Germany, UK and Australia, to assess to
what extent a gap in reported SWB of the spouses affects the likelihood of divorce.
Using fixed effect estimation techniques, they find that indeed a higher satisfaction gap
8
gives a higher likelihood of partnership dissolution - and these results are robust to a
range of specifications - and across the three longitudinal surveys considered. Their
argument is that when it comes to assessing the role of one's own SWB with respect to
objective measureable outcomes, such as a divorce, the key reference group is in fact
that of the spouse. The analysis of Guven et al. (2012) also uncover other interesting
insights. For instance, the positive effect of the SWB gap on divorce, is not only driven
by any deviation away from the baseline case at the time of partnering, but also that the
absolute level of the gap matters. Secondly, the effects are potentially asymmetric, and
for divorce, they find that its likelihood increases especially when the wife has a lower
level of SWB than the husband, but not the other way around.
To the best of our knowledge there are so far no studies considering the effect of
partners' joint SWB on childbearing. For our analysis, the study by Le Moglie et al.
(2015) gives important clues to what expect. First, one may expect different effects
depending on parity. Secondly, there might be gender dominance when deciding to have
children. In line with Testa et al. (2011), we are interested in understanding if the SWB
of the female partner has a stronger impact on the childbearing decision than the male
partner. The research questions of this study are consequently summarized as follows:
1) Does the subjective well-being of one partner, either the female or male, prevail over
the other in the decision of having a(another) child? 2) Is there a multiplicative effect of
the subjective well-being of the two partners in the decision of having a(nother) child?
Building on the findings of Testa et al. (2011), that shows that women have a greater
influence on fertility decisions than men, we postulate that SWB of the female partner
plays a stronger role than male partner’s SWB, though a priori we do not have a
specific hypothesis regarding differential effects for parities. We moreover pose another
9
hypothesis about the multiplicative effect of both partners’ happiness. Specifically we
test whether having a child is more likely in couples where both partners have a high
level of SWB compared with couples in which only one partner has a high SWB.
3. Data and methods
3.1. Sample selection and measurements
We use the British Household Panel Survey (BHPS), an annual panel survey consisting
of a nationally representative sample of about 5,500 households recruited in 1991,
containing a total of approximately 10,000 interviewed individuals. Participants are re-
interviewed each successive year for 18 years and, if they split from original households
to form new households, they are followed and all adult members of these households
are also interviewed. Similarly, new members joining sample households become
eligible for interview and children are interviewed as they reach the age of 16. The
BHPS dataset is well-suited to investigating the relationship between happiness and
fertility because it provides information on several socio-economic characteristics,
fertility history, and subjective well-being measured over time.
We select only observations for couples, either married or cohabiting, which
means that we exclude from the analyses observation-years where individuals where
observed as single, divorced or widowed and we also excluded partnered individuals for
which the information on the partner was missing. We select all couple-year
observations for heterosexual couples where the woman is aged 16-45 and the man is
aged 16-50. To allow the effect of happiness to differ by parity we considered
separately the transition to the first, second and higher order births. After deletion of
10
missing values and a few cases of multiple births our working samples consists of 645,
554 and 264 couples for the analyses of the first, second and higher order births
respectively, corresponding to 2101, 1706 and 982 couple-year observations. Of course
the same couple could experience more than one transition during the observation
period, whereas others may have entered the survey with already one or more children.
The key explanatory variables measure women and men’s happiness. The BHPS
questionnaire ask: Have you recently been feeling reasonably happy, all things
considered?". Possible responses are: more than usual, same as usual, less so and much
less. This question is asked in each wave of the survey and therefore it was preferred to
the question on life satisfaction that is missing in 6 waves of BHPS (wave 1 to 5 and
wave 11). As reported by Myrskylä and Margolis (2014), happiness and life satisfaction
are highly correlated and offer consistent results. Since the percentage of respondents
who declared to feel "much less happy than usual" was extremely low (< 3% for both
men and women), we decided to group this and the "less so" categories. We introduced
two categorical variables in our regression models measuring women and men reported
happiness separately: women (men) "happier than usual" and "less happy than usual".
The reference category is "as happy as usual".
To test for possible interaction between partners' happiness, in a second analysis
we built 9 couple types based on the combination of the happiness levels of both
partners: both man and woman less happy than usual; man less happy than usual and
women at the usual level and so on (reference: both partners at the usual level).
Keeping in line with the existing literature (Margolis and Myrskylä, 2015;
Myrskylä and Margolis and 2014; Pollmann-Shult 2014), we introduced a set of control
variables. We introduced a dummy variable indicating whether the couple is cohabiting
11
or married (Keizer and Schenk 2012). All the other control variables refer to
individuals' characteristics and are measured for both partners separately. In particular,
we controlled for both partners’ age, health, education and working status. Age and
health are introduced as numerical variables. Health is measured by self-rated health on
a five-point scale (higher scores indicate worse health). Working status is introduced
trough a set of two dummy variables (inactive, unemployed), “employed” being the
reference category. Education is measured by a categorical variable indicating the
highest level of education attained by the individual, that is, degree (reference category),
diploma, vocational school, lower school level.
3.2. Empirical approach
We analyze the transition to the first (and higher order births) using discrete-time event
history logit models (see e.g., Allison, 1982). Formally, it is assumed that time takes on
positive integer values (t = 1, 2, 3, …), we observe n independent couples (i = 1, …, n)
and the observation continues until time ti, at which point either a childbirth event
occurs or the observation is censored. We model the (discrete-time) hazard rate of
experiencing a childbearing event within the time interval t, pit, as follows:
( ) ( ) ( ) ( )( ) ,happierwomen
happy lesswomen happiermen happy lessmen plogit
4
3210it
iitit
itititt
X ηγβ
ββββα
+++
++++= (1)
12
where αt represents the logit of the baseline hazard function, that we specify as a
quadratic function of time, and β1 to β4 are the effects of our interest. Again, it is worth
noting that the reference category consists of couples who report the same level of
happiness. X is the set of control variables listed above (including both time-variant and
time-invariant covariates) and ηi represent the couple-level unobserved heterogeneity
modeled as a normal variable with zero mean and variance to be estimated.
We estimate model (1) by parity. Specifically, we consider three analyses: for
the transition to the first, second and higher order parities. For example, in the first case
couples enter the risk-set the first wave they are surveyed if childless and they are
followed till they have the first child or exit the survey. For the transition to parities
higher than the second, in principle we can observe repeated events and each time a
couple experiences a childbirth event it re-enters the set at risk of another event from the
following time point.
To test the multiplicative effect of partners' happiness we also estimated
discrete-time event history logit models including the couples’ happiness types
described above instead of the four explanatory variables showed in equation (1).
4. Results
4.1. Descriptive statistics
Table 1 presents descriptive statistics on the independent variables. In particular, for
each of the three samples used in the multivariate analyses (that correspond to the three
samples described above) we calculate the percentage in each category of the
13
independent variables in the year preceding the birth of a child. For the continuous
variables we present the average values. We observe that both the majority of men and
women declare to be as happy as usual. Both the percentage of women that are less
happy and happier than usual tend to be higher than the corresponding percentages for
men.
[INSERT TABLE 1 ABOUT HERE]
As for the control variable, we observe that the percentage of cohabiting couples
reduces from 37.4% in the first sample (couples at risk of the first child) to 21.4 in the
third sample. Both men and women are, on average, in good health (around 2 points).
While the vast majority of men are employed in all three samples, not surprisingly the
percentage of employed women decreases considerably from 74.3% in the first sample
to 44.8% in the third sample.
Table 2 provides us with an overview of how many couples have partners with
similar or different happiness level and shows which couple’s types are more common.
In this case, for the sake of brevity, we averaged over all observations and parities. The
most common type (46.2% of all the couples) is the one in which both partners are at
the usual level of happiness, followed by those couples in which either the man or the
woman is happier than usual while the partner is as happy as usual (11% and 14.6%,
respectively). There are also quite a few couples where one of the partners is less happy
than usual and the other partner's happiness is at the usual level. Other couple types are
less prevalent. Given the presence of "discordant" couples, the association between
14
partners' happiness is high but not too strong to make us worried about multi-
collinearity (the Kendall's tau rank correlation coefficient is 0.14). However, estimating
the association between some of the 9 partners' happiness combination and fertility may
be hampered by low N in some of the cells of table 2.
[INSERT TABLE 2 ABOUT HERE]
4.2. Discrete-time event history logit models results
Table 3 reports estimates of discrete-time event history logit models predicting
childbirth transitions as function of both partners’ happiness, our explanatory variables
of interest, and a set of control variables.
[INSERT TABLE 3 ABOUT HERE]
Estimates of the first model in Table 3 show that either when the man or the
woman is happier than usual, the probability of having a first child increases (as with
respect to when he or she is at the usual happiness level) and the effect is strongly
significant. Moreover, the effect is greater for women than for men. When considering
transitions to second or higher order parities, we do not find any significant effect of
men’s happiness while for women we find that both a lower and a higher happiness
level than the usual one is associated with a smaller probability of transition to the
15
second child. For transitions to higher order parities we do not find any statistically
significant effect of happiness.
Table 4 present estimates of discrete-time event history logit models where we
test for interactions between partners' happiness. We do so by introducing dummy
variables for couples types based on the combination of partners' levels of happiness
(the reference being both partners at the usual level of happiness).
Estimates in the first column show that three types of couples have significantly
higher probabilities to have a first child than couples where both partners are as happy
as usual. This includes couples where one of the partners reports an above-average level
of happiness, while the other is at the usual level. However, we notice that when the
woman is happier, the positive effect on the probability to have the first child is bigger
than the case where the man is happier than usual. We also find support for a
multiplicative effect of partners’ happiness: when both partners are happier than usual
the positive effect on fertility is stronger than when only one of the two is happier, while
the other partner's happiness is at the usual level.
For transitions to the second child, we notice that the two “extreme” types of
couples, i.e. those where both partners report a level of happiness below or above the
usual level, show significant lower probabilities to have a second child with respect to
the reference couple (where both partners are at the usual level of happiness).
Interestingly, the estimated coefficients are very similar (and not significantly different).
Also couples where the woman is happier than usual while the man is as happy as usual
are less likely to progress to the second parity.
16
For higher order transitions, consistently with the results in Table 3, we do not
significant differences among the different couple’s types.
[INSERT TABLE 4 ABOUT HERE]
5. Conclusions
The aim of the paper is to bring further insights into the relationship between subjective
wellbeing and childbearing decision making by taking a couple perspective. Existing
literature on the relationship between SWB and fertility has taken the respondent as the
unit of analysis, neglecting that the decision to have a child is a couple decision. We
uncover whether there is a gender dominance when deciding to have children. Moreover
we explore to what extent a potential coherency or mismatch in subjective wellbeing
across partners may affect their fertility decision making. By doing so we investigated
to what extent there are multiplicative effects of the two partners’ subjective wellbeing,
that is a stronger effect on childbearing if both partners report consistent happiness
level. We also implement our analysis by parity, since the subjective wellbeing of the
two partners as well as their mismatch or coherency may differently affect the decision
to have a first child or another one.
We find that a higher level of happiness of both men and women is associated
with a higher risk of having a first child, meaning that happier people are more likely to
become parents sooner. On the other hand, women's happiness seems to matter more
17
strongly than that of the male partner in the decision to have a second child. Specifically
we observe that both high and low levels of women’s happiness are associated with a
lower risk of having a second child. This latter finding implies that women may not
want to have another child not to change the positive status-quo in which they live,
suggesting an aversion to lifestyle changes (Mencarini and Tanturri, 2007). All of these
findings are relevant in showing that there is not a one sided and general gender
dominance in the fertility decision making when the SWB is concerned - instead it
depends on the parity.
Another key finding is that there is a multiplicative effect of the SWB of both
partners on the decision to have a child. We indeed observe that when both the woman
and the man report a particularly high level of happiness the probability of becoming
parents for the first time increases more than when only one of the two partners is
happier than usual. Another multiplicative effect is found after couples had the first
child: if both parents are happier than usual, this time the risk of having a second child
is the lowest. Put another way, a coherent and high level of SWB of both partners leads
to the highest (or lowest) probability for the couple to have a(nother) child, depending
on the parity This is an important result because it shows that an individual perspective
may provide only a partial understanding of the SWB-fertility relationship.
Although we acknowledge that our analysis cannot uncover strict causal effects
of parents’ SWB on fertility, we believe this paper expands existing literature by
adopting a novel couple approach in studying how happiness levels of both partner prior
to childbearing are associated with future fertility outcomes.
18
We encourage further research to investigate further the way couple dynamics
could affect partners’ wellbeing and in turn fertility choices.
19
References Aassve, A, Mazzuco, S., & Mencarin,i L. (2005). Childbearing and wellbeing: A
Comparative Analysis of Welfare Regimes in Europe. Journal of European
Social Policy, 15(3), 283- 299.
Aassve, A., Goisis, A., & Sironi, M. (2012). Happiness and Childbearing across Europe.
Social Indicator Research, 108(1), 65-86.
Aassve, A., Sironi, M., & Bassi, V. (2013). Explaining Attitudes Towards Demographic
Behaviour. European Sociological Review, 29(2), 316-333.
Aassve, A., Fuochi, G., Mencarini, L., & Mendola D. (2015). What is your couple
type? Gender ideology, housework sharing, and babies. Demographic Research,
32, 835-858.
Allison, P. D. (1982) Discrete-time methods for the analysis of event histories.
Sociological methodology, 13(1), 61-98.
Balbo, N., & Arpino, B. (2014). The role of family orientations in shaping the effect of
fertility on subjective well-being. Swellfer WP n.4, Collegio Carlo Alberto, Italy.
Baranowska, A., & Matysiak, A. (2011) Does parenthood increase happiness? Evidence
for Poland. Vienna Yearbook of Population Research, Vol. 9, (2011), pp. 307-
325.
Becker, G. S., Landes, E. M., & Michael, R. T. (1977). An Economic Analysis of
Marital Instability. Journal of Political Economy, 85, 1141- 87.
20
Bauer, G., & Kneip, T. (2013). Fertility from a couple perspective: A test of competing
decision rules on proceptive behaviour. European Sociological Review, 29(3),
535-548.
Billari, F. C. (2009). The Happiness Commonality: Fertility Decision in Low-Fertility
Settings. In UNECE (Ed.), How Generations and Gender Shape Demographic
Change (pp. 7-38). New York and Geneva: United Nations.
Clark, A.E., Diener, E., Georgellis, Y., Lucas, R.E. (2008). Lags and leads in life
satisfaction: a test of the baseline hypothesis. Economic Journal, 118 (529),
222–243.
Clarkwest A. (2007). Spousal dissimilarity, race, and marital dissolution. Journal of
Marriage and Family, 69, 639–653
Greenwood, J., Guner N., Kocharkov, G., & Santos C. (2014). Marry Your Like:
Assortative Mating and Income Inequality. American Economic Review, 104(5),
348–53.
Guven, C., Senik, C., & Stichnoth, H. (2012). You can't be happier than your wife:
Happiness gaps and divorce. Journal of Economic Behavior & Organization,
82(1), 110–130.
Inglehart, R. (1971). The Silent Revolution in Europe: Intergenerational Change in
Post-Industrial Societies, American Political Science Review, 65(4), 991–1017.
Jalovaara, M. (2003). The joint effects of marriage partners' socioeconomic positions on
the risk of divorce. Demography, 40(1): 67-81.
21
Keizer, R., & Schenk, N. (2012). Becoming a parent and relationship satisfaction: A
longitudinal dyadic perspective. Journal of Marriage and Family, 74(4), 759-
773.
Kohler, H., Behrman, J. R., & Skytthe, A. (2005). Partner + children = happiness? The
effects of partnerships and fertility on well-being. Population and Development
Review, 31(3), 407– 445.
Kravdal, O. (2014). The Estimation of Fertility Effects on Happiness: Even More
Difficult than Usually Acknowledged. European Journal of Population, doi
10.1007/s10680-013-9310-9.
Le Moglie, M., Mencarini, L., & Rapallini, C. (2015). Is it just a matter of personality?
On the role of life satisfaction in childbearing behavior. Journal of Economic
Behavior and Organization, forthcoming.
Margolis, R., & Myrskylä M. (2011). A Global Perspective on Happiness and Fertility.
Population and Development Review, 37(1), 29-56.
Margolis, R., & Myrskylä, M. (2015). Parental well-being surrounding first birth as a
determinant of further parity progression. Demography, 52(4), 1147-1166.
McDonald, P. (2002). Sustaining fertility through public policy: The range of options.
Population-E 57(3): 417-446.
Mencarini, L., & Tanturri, M.L. (2007). High fertility or childlessness: micro-level
determinants of reproductive behaviour in Italy. Population, 61(4), 389–415
22
Milewski, N and Kulu, H. (2014). Mixed marriages in Germany: a high risk of divorce
for immigrant-native couples. European Journal of Population, 30 (1), 89-113.
Myrskylä, M., & Margolis, R. (2014). Happiness: before and after the kids.
Demography, 51(5), 1843-1866.
Parr, N. (2010). Satisfaction with life as an antecedent of fertility. Demographic
Research, 22(21), 635-662.
Perelli-Harris, B. (2006). The influence of informal work and subjective well-being on
childbearing in post-Soviet Russia. Population and Development Review, 32(4):
729-753.
Pollmann-Shult, M. (2014). Parenthood and Life Satisfaction: Why Don’t Children
Make People Happy? Journal of Marriage and Family, 76, 319–336.
Powdthavee, N. (2009). I can’t smile without you: Spousal correlation in life
satisfaction. Journal of Economic Psychology 30, 675–689.
Testa, M.R., Cavalli, L., & Rosina, A. (2011), Couples' childbearing behaviour in Italy:
which of the partners is leading it? Vienna Yearbook of Population Research,
157-178.
Van de Kaa, D. J. (1987). Europe’s Second Demographic Transition, Population
Bulletin, 42 (1), Washington, The Population Reference Bureau.
23
Tables Table 1. Descriptive statistics (%) on independent variables for each of the three sample used in the multivariate analyses.
Variable First child Second child Third+ child Man’s happiness
Less happy than usual 9.4 10.4 10.3 As usual 69.0 69.9 68.9 Happier than usual 21.7 19.7 20.8 Woman’s happiness
Less happy than usual 12.9 12.1 15.2 As usual 58.6 63.7 66.8 Happier than usual 28.4 24.2 18.0 Coabithing couple 37.4 25.2 21.4 Man's age (mean) 30.1 31.0 32.2 Woman's age (mean) 28.0 28.7 29.4 Man's health (mean) 1.9 1.9 2.1 Woman's health (mean) 2.0 2.0 2.1 Man's education level
Degree 21.2 20.1 13.2 Diploma 24.9 26.7 24.9 Vocational school 36.2 33.3 33.5 Low school level 17.7 19.9 28.4 Woman's education level
Degree 19.6 18.2 15.9 Diploma 29.4 27.8 22.5 Vocational school 39.0 41.4 39.1 Low school level 12.0 12.6 22.5 Man's working status
Inactive 2.5 2.6 4.2 Unemployed 5.7 5.9 12.1 Employed 91.8 91.5 83.6 Woman's working status
Inactive 21.9 34.5 52.8 Unemployed 3.7 2.7 2.4 Employed 74.3 62.8 44.8 N. couple-observations 2101 1706 982 N. couples 645 554 264
24
Table 2. Combination of women's and men's happiness, average values over all observations.
Men's happines Women's happines
Less happy As usual Happier Less happy 2.5 6.6 1.5 As usual 9.0 46.2 14.6 Happier 2.0 11.0 6.7
25
Table 3. Discrete-time event history logistic regression estimates for childbirth transitions as function of both partners' happiness, by parity.
Independent variables First child Second child Third+ child Man’s happiness (ref.: as usual)
Less happy than usual 0.013 -0.192 -0.137
(0.277) (0.206) (0.265)
Happier than usual 0.396** 0.006 -0.117
(0.190) (0.155) (0.208)
Woman’s happiness (ref.: as usual) Less happy than usual -0.197 -0.396** -0.014
(0.255) (0.197) (0.226)
Happier than usual 0.596*** -0.502*** 0.005
(0.175) (0.151) (0.227)
Coabithing (ref.: married) -0.866*** -0.515*** -0.429*
(0.210) (0.185) (0.257)
Man's age -0.479*** 0.171 0.270
(0.181) (0.120) (0.181)
Woman's age 0.290 -0.175 -0.195
(0.181) (0.139) (0.202)
Man's age squared 0.008*** -0.002 -0.003
(0.003) (0.002) (0.003)
Woman's age squared -0.003 0.004 0.004
(0.003) (0.002) (0.003)
Man's health 0.024 -0.028 -0.164
(0.100) (0.074) (0.107)
Woman's health -0.031 0.084 0.041
(0.094) (0.071) (0.104)
Man’s Education (ref.: degree) Diploma 0.413 0.202 0.579*
(0.273) (0.193) (0.300)
Vocational school 0.315 0.257 0.427
(0.270) (0.194) (0.301)
Low school level 0.842** 0.205 0.708**
(0.334) (0.221) (0.323)
Woman’s Education (ref.: degree) Diploma 0.348 0.274 0.485*
(0.276) (0.202) (0.287)
Vocational school 0.602** 0.076 0.282
(0.280) (0.195) (0.276)
Low school level 0.519 0.136 0.358
(0.363) (0.246) (0.320)
Employment (ref.: employed) Inactive man -0.552 0.610 0.496
(0.575) (0.377) (0.399)
Unemployed man 0.053 0.112 0.444*
(0.348) (0.249) (0.249)
Inactive woman 4.294*** 1.432*** 0.951***
26
(0.322) (0.130) (0.185)
Unemployed woman 0.761* -0.010 -0.566
(0.403) (0.561) (0.802)
t -0.026 -0.214*** -0.080
(0.070) (0.050) (0.065)
t squared 0.009** 0.015*** 0.008**
(0.004) (0.003) (0.004)
Constant -1.891 -2.711 -5.293
(2.811) (2.209) (3.221)
N. couple-observations 2101 1706 982 N. couples 645 554 264
Note: * p-value < 0.10; ** p-value < 0.05; *** p-value < 0.01.
27
Table 4. Discrete-time event history logistic regression estimates for childbirth transitions as function of couples happiness types, by parity.
Independent variables First child Second child Third+ child Couple’s happiness (ref.: man and woman at the usual level)
Man less, Woman less -0.167 -0.652*** -0.370
(0.337) (0.234) (0.250)
Man usual, Woman less 0.113 -0.378 -0.193
(0.308) (0.240) (0.276)
Man less, Woman usual 0.381 0.029 -0.313
(0.348) (0.244) (0.337)
Man more, Woman less 0.024 0.590 0.147
(0.597) (0.444) (0.512)
Man less, Woman more 0.502 -0.352 -0.618
(0.608) (0.501) (0.806)
Man more, Woman usual 0.564** -0.002 -0.139
(0.270) (0.196) (0.250)
Man usual, Woman more 0.727*** -0.401** 0.101
(0.216) (0.179) (0.277)
Man more, Woman more 0.987*** -0.711*** -0.323
(0.275) (0.269) (0.418)
Coabithing (ref.: married) -0.863*** -0.470*** -0.255
(0.207) (0.178) (0.239)
Man's age -0.487*** 0.204* 0.158
(0.178) (0.114) (0.166)
Woman's age 0.292 -0.202 -0.032
(0.180) (0.131) (0.188)
Man's age squared 0.008*** -0.003 -0.001
(0.003) (0.002) (0.002)
Woman's age squared -0.003 0.004* 0.001
(0.003) (0.002) (0.003)
Man's health 0.022 -0.039 -0.139
(0.100) (0.071) (0.098)
Woman's health -0.040 0.087 0.027
(0.094) (0.068) (0.098)
Man’s Education (ref.: degree) Diploma 0.372 0.209 0.523*
(0.279) (0.190) (0.293)
Vocational school 0.270 0.258 0.390
(0.274) (0.191) (0.294)
Low school level 0.805** 0.229 0.574*
(0.324) (0.211) (0.310)
28
Woman’s Education (ref.: degree) Diploma 0.357 0.286 0.346
(0.280) (0.197) (0.276)
Vocational school 0.557** 0.109 0.109
(0.282) (0.191) (0.264)
Low school level 0.555 0.148 0.236
(0.352) (0.236) (0.301)
Employment (ref.: employed) Inactive man -0.638 0.443 0.346
(0.544) (0.338) (0.351)
Unemployed man 0.033 0.111 0.520**
(0.346) (0.241) (0.233)
Inactive woman 4.365*** 1.396*** 0.997***
(0.316) (0.126) (0.177)
Unemployed woman 0.872** 0.132 -0.456
(0.400) (0.510) (0.795)
t -0.005 -0.226*** -0.067
(0.070) (0.048) (0.062)
t squared 0.008** 0.016*** 0.007*
(0.004) (0.003) (0.004)
Constant -1.848 -2.775 -5.820*
(2.765) (2.135) (3.096)
N. couple-observations 2101 1706 982 N. couples 645 554 264
Note: * p-value < 0.10; ** p-value < 0.05; *** p-value < 0.01.